-
The African Statistical Journal, Volume 3, November 2006 111
1: Principal Statistician, Uganda Bureau of Statistics, P.O.Box
7186, Kampala, Uganda [email protected]
The Role of Household Surveys in Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
James Muwonge1
Summary
Uganda’s Poverty Eradication Action Plan (PEAP) and the
Millennium Development Goals (MDGs) have put a lot of focus on the
statistical systems that must produce the data to monitor the PEAP
and the MDGs indicators. The focus and direction of the data
requirements in Uganda has greatly evolved and prompts government
and other key stakeholders in the development process to establish
a sustainable moni-toring system of interventions. The PEAP and the
MDGs among other plans require quality and comprehensive data for
regular progress monitoring and review. Nation-ally representative
household surveys undertaken by the Uganda Bureau of Statis-tics
(UBOS) have contributed greatly to the policy debate and
development in Ugan-da. The paper provides the UBOS’s experience in
meeting the various data demands at the national and international
levels through this survey programme..
Keywords
Poverty Eradication Action Plan, Household Survey Programme,
Qualitative and Quantitative approaches to poverty monitoring, the
long term household survey pro-gramme.
Résumé
Le Plan d’Action d’Eradication de la Pauvreté (PAEP) de
l’Ouganda et les Objectifs du Millénaire pour le Développement
(OMD) ont mis une grande priorité sur les sys-tèmes statistiques
qui doivent produire des données pour le suivi des indicateurs du
PAEP et des OMD. La priorité et l’orientation de la demande des
données en Ougan-da ont fortement évolué et ont incité le
gouvernement et d’autres principales parties prenantes dans le
processus de développement à établir un système durable du sui-vi
des interventions. Le PAEP et les OMD parmi tant d’autres plans
requièrent des données exhaustives et de bonne qualité pour faire
le point et un suivi régulier des progrès accomplis. Des enquêtes
auprès des ménages, representatives au plan na-
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006112
tional, entreprises par le Bureau Ougandais de Statistique
(UBOS) ont contribué lar-gement au débat politique et au
développement en Ouganda. L’article présente l’ex-périence de
l’UBOS pour satisfaire diverses demandes de données aux niveaux
national et international grâce à ce programme d’enquêtes.
Mots clés
Plan d’Action pour l’Eradication de la Pauvreté, Programme
d’Enquêtes auprès des Ménages, Approches quantitatives et
qualitatives pour le suivi de la pauvreté, Le pro-gramme de long
terme d’Enquêtes auprès des Ménages.
1. Background
Discussions of the need for reliable data for policy-making have
reached an almost unprecedented level at both national and
international levels. Uganda’s Poverty Erad-ication Action Plan
(PEAP) - the national development framework, and the Millenni-um
Development Goals (MDGs) have focused on statistical systems that
must pro-duce data to monitor the PEAP and the MDGs indicators.
Utilization of data for planning is not new in Uganda but most
of the information used in the past was mainly from administrative
sources. These included financial and economic data series that
were and continue to be produced as part of the routine record
keeping. Since 1997, Uganda has undertaken many reforms that have
wide-ly affected the economy and the welfare of Ugandans in
general. Overtime, the de-mand for data has also evolved in line
with these changes. The demand has shifted from being simple
counting exercises to complex programs designed to determine the
causes of individual and household behavior and the effect of
government poli-cies on the population’s choices and welfare.
Indeed, the current statistical systems in the world represent the
result of centuries of actions, research, and methodologi-cal
improvements in the world’s ability to understand the interaction
of various forc-es, on the socio-economic welfare of individuals,
households, and countries. Appar-ent improvements in data must take
into account the existing systems, their users, and the traditional
data and emerging needs therein.
The importance of household survey statistics for national
development cannot be over emphasized as it is needed to indicate
among other things the welfare lev-el to guide policy makers in
framing socio-economic developmental plans and initi-ate
interventions for improving people’s socio-economic conditions. The
discussion
-
The African Statistical Journal, Volume 3, November 2006 113
of the need for more and better data for policy making has
reached an almost un-precedented level at both national and
international levels. Uganda’s PEAP and the MDGs have greatly
focused on the statistical systems that must produce the data to
monitor the PEAP, MDGs and other development indicators. The Uganda
Bureau of Statistics (UBOS) has contributed to both the
understanding of various phenomena through the household survey
programme and specifically to the debate on the ef-fects on welfare
levels of programmes on the population.
The availability of information on a regular basis is important
in evaluating the dynam-ics of various economic factors which occur
as a result of economic development. Information from the household
surveys have provided the foundation and monitoring mechanism for
the poverty eradication efforts in Uganda. For instance, Uganda has
been able to monitor and measure trends of poverty, and changes
that occur among the different population groups using the
comprehensive datasets from national-ly representative household
surveys. According to the information from household surveys, the
percentage of the population below the poverty line decreased from
56 percent in 1992 to 44 percent in 1997 and further to 38 percent
in 2003. Such in-formation collected through household surveys has
influenced the Poverty Eradica-tion Action Plan (PEAP) revision
process in Uganda.
2. The National Development Framework
The Poverty Reduction Strategy Papers (PRSPs) have been used by
the World Bank (WB) and International Monetary Fund (IMF) as a
pre-condition for development assistance to low income countries to
qualify for external financing and debt re-lief. The PEAP is
Uganda’s development framework and was first drafted in 1997 and
revised in 2000 and 2003/04. The PEAP is a dynamic document which
is re-vised every four years in accordance to the changing
circumstances and emerging priorities within the national economy.
Data and information from household surveys have provided
substantial inputs to the revision processes. For instance, the
2002-2004 PEAP revision was influenced by four core challenges,
namely: highly unequal growth leading to increased poverty (based
on household survey data), less than expected improvement in human
development indicators in the 1990s (with the ex-ception of
HIV/AIDS), and persistent insecurity which resulted into changes in
the regional pattern of poverty. Based on these challenges, the
PEAP 2004 has five pillars/components namely: a pillar focusing on
Economic management; Produc-
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006114
tion, competitiveness and incomes; Security, conflict-resolution
and disaster-man-agement; Governance and Human development. In
addition, information from house-hold surveys has substantially
contributed to the monitoring of MDGs.
3. The Status of Household Surveys Prior to the PEAP
Household surveys were only revived in 1988 after a period of
almost 20 years. In 1990, poverty alleviation programmes were
started under the Programme for the Al-leviation of Poverty and the
Social Cost of Adjustment (PAPSCA). Under the Social Dimensions of
Adjustment component of PAPSCA, data on a wide range of social and
economic statistics was collected with the Integrated Household
Survey (IHS) providing the first baseline data on many of the
issues. Thereafter, a proposal to de-velop a clear programme of
poverty alleviation was developed and endorsed by the Ministry of
Finance, Planning and Economic Development. Because of the many and
varied data gaps, the surveys conducted after 1992 were attempts to
fill gaps. The modules covered in these household surveys were
based on discussions with some of the key users of data and these
discussions culminated into the first PEAP in 1997. The uncertainty
about availability of funds to undertake regular surveys was the
main challenge faced during the post PAPSCA era. An overview of the
situation before and after PEAP is summarized in the Table 1
below.
Table 1: Pre and post PEAP data appreciation
Period before PEAP After PEAP (1997+)
Limited use and availability of household survey data Increased
demand for and use of household survey data and information
Initiated household surveys under Programme for alleviation of
Poverty and the Social Cost of Adjust-ment (PAPSCA)
The PEAP is the building block for the household survey
programme
Limited users and producer consultations Extensive consultation
with government, development partners and other stakeholders
Data generated almost exclusively for government and a few
development partners
Users of data are many and varied
Virtually no collaboration with research institutions There is
extensive collaboration with research institu-tions and increased
production of policy relevant papers based on household survey
data
Very limited analysis of household survey data Increased access
to household survey Data
-
The African Statistical Journal, Volume 3, November 2006 115
Access to data limited Data collected to monitor PEAP and other
development frameworks
The main focus was on macroeconomic variables (GDP, Consumer
Price Indices etc)
The focus is on both macro and micro indicators
Data generated to fill gaps due to over 20 years without
household survey data
Increased level of analysis and appreciation of the qual-ity of
data
4. Stakeholders in the Poverty Monitoring Process in Uganda
During the 1990s, poverty reduction became a priority in many
countries mainly sup-ported by development partners. The outcome of
the currently widespread notion of poverty reduction was the
emergence of Poverty Reduction Strategy Papers (PRSP) or the PEAP
in case of Uganda, and the need to monitor the performance of
pover-ty related interventions. The PEAP outlines the monitoring
and evaluation mecha-nisms through which the outcomes would be
updated. Monitoring and Evaluation is important in keeping on track
development programmes and provides the relevant information to
decision makers for making informed decisions. It also keeps other
stakeholders, namely the legislature, the public, civil society
organizations and de-velopment partners, informed about the
progress being made in implementing the PEAP and other development
frameworks.
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006116
The monitoring system in Uganda is multi faceted having many
institutions and agen-cies running independent monitoring and
evaluation systems. This has resulted into inconsistent reporting,
duplication of efforts and wastage of resources. These some-times
uncoordinated systems have undermined the overall goal of informing
the pov-erty eradication programme. An integrated programme has
been developed under the PEAP to minimize these constraints and
harmonise the reporting mechanism. Poverty monitoring is a
responsibility of many stakeholders ranging from UBOS to Civil
Society Organizations. Table 2 presents a summary of the
stakeholders and their respective roles in poverty monitoring in
Uganda.
Table 2: Institutional framework for poverty monitoring in
Uganda1
Institution Responsibility
Office of the Prime Minister Collate policy related information
and influence national political, socio-economic decisions.
Ministry of Finance, Planning and Economic Development:• Poverty
Monitoring and Analysis Unit.• Budget Department• Macroeconomic
Dept.
Coordinate poverty data collection particularly the Powerty
Participatory Assesments. Analyze the whole range of data. Publish
poverty reports that present a holistic view of poverty in Uganda
and the implementa-tion of the PEAP. Disseminate findings to
Government and civil society, and service budget working groups.
Commission poverty research and evaluations.Monitor Public
ExpenditureMonitor and project macroeconomic indicators.
Uganda Bureau of Statistics Conduct censuses and surveys
(including National Integrated Household Surveys, National Service
De-livery Surveys, and Demographic and Health Surveys) and provide
basic analysis of these data. Construct National Accounts
Ministry of Public Service Provide information on public service
performance (Results Oriented Management-ROM).
Sectoral ministries, particularly the Management Information
Systems in Planning Units.
Design indicators and collect administrative data on service
delivery efforts and their immediate outcomes; analyze these data
in conjunction with other data sources, and identify policy
responses.
Uganda AIDS Commission Coordinate data and policy response on
AIDS.
Police, Judiciary, Prisons, other JLOS Institutions Compile
administrative data on crime and the activities of the police and
judiciary.
Office for Co-ordination of Humanitarian Assistance Co-ordinate
and publish data on the number and living conditions of displaced
people and refugees.
Inspector General of Government, Auditor-General,Dept. of Ethics
and Integrity
Ensuring that inputs are converted into outputs in a transparent
manner by monitoring the integrity of public expenditures
1: Adopted from Margaret Kakande: Poverty Monitoring in Uganda
2006: The Practices and Emerging Issues, April 2006
NGO – Non-governmental Organization, CSO – Civil Society
Organization
-
The African Statistical Journal, Volume 3, November 2006 117
District Authorities Collect information on relevant outputs,
inputs, and outputs at this level for their own planning purposes,
and sharing with central government.
Economic Policy Research Centre Conduct relevant economic policy
research that informs the monitoring process, thereby helping in
refining it further.
Academic institutions and NGOs
• Uganda Human Rights Commission
Conducting research on all aspects of poverty, using official
data and collecting their own, providing an independent view on
poverty.Providing information on human rights issues
NGOs/CSOs Undertake value for money evaluation activities
Development Partners Supporting the monitoring efforts of
government and CSOs; and using the findings to positively influence
poverty reduction activities.
The Poverty Monitoring Strategy uses three main types of data;
quantitative data collected by the Uganda Bureau of Statistics
(UBOS) through censuses and sur-veys, those collected by sector
ministries through their Management Information Systems, and those
collected through the Participatory Poverty Assessment Pro-grammes
(PPAs).
5. The Role of National Household Survey Data in PEAP
Monitoring
During the 1970s and the early 1980s, there was great dearth of
data on house-holds in general in Uganda. This was a period when
data collection at all levels (household, establishments, etc) was
in limbo. Since 1988, however, the need for evidence based
decision-making emphasised by the country’s planning and decision
making authorities and the political leadership, culminated into
the rebirth of dynam-ic statistical programmes at UBOS and in
Uganda in general. Since then, Uganda has undertaken more household
surveys than many Sub-Saharan African countries in the last 15
years. Table 5.1 provides a summary of the Household Surveys
under-taken since 1988.
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006118
Table 3: Household surveys conducted by UBOS from 1988 to
2004
From the standpoint of their ability to address the policy
issues emerging from the PEAP and other development frameworks, the
variety of surveys can be grouped into two major categories namely
Consumption Surveys (developed under different names such as HBS,
IHS, MS, and, more recently, UNHS,) and Demographic and Health
Surveys (DHS and Sero-prevalence Surveys). Below is the summary
descrip-tion of the objectives of each of the consumption
surveys;
5.1 Consumption surveys (household surveys with a consumption
module)
The Household Budget Survey (HBS), conducted from April 1989 to
March 1990 was the first detailed household expenditure-based
survey since the 1960s. The main objective of the survey was to
provide basic data needed to revise the Consum-er Price Index and
data for improving estimates of the household final consumption
expenditure component of the Gross Domestic Product (GDP) using the
expenditure approach. As outlined in Table 3, the first large scale
Integrated Household Survey (IHS) conducted in 1992 was a
multi-subject inquiry with the main objective of pro-viding a
complete dataset needed to understand the mechanisms and effects of
the structural adjustment programs, to fill the socio-economic data
gaps and to provide base-line data relating to key economic
indicators used in development planning.
Uganda
Demogra-
phic and
Health
Survey
Household
Budget
Survey
Integrated
Household
Survey
Uganda
National
Household
Survey
1993-94
Uganda
National
Household
Survey
1994-95
Uganda
Demogra-
phic and
Health
Survey
Uganda
National
Household
Survey
1995-96
Uganda
National
Household
Survey
1997
Uganda
National
Household
Survey
1999-00
Uganda
Demogra-
phic and
Health
Survey
Uganda
National
Household
Survey
2002-03
National
Service
Delivery
Survey
(NSDS)
UDHS HBS HIS MS-1 MS-2UDHS
(+MICS)MS-3 MS-4 UNHS I
UDHS
(+MICS)UNHS II NSDS
From Sep-88 Apr-89 Feb-92 Aug-93 Jul-94 Mar-95 Sep-95 Mar-97
Aug-99 Sep-00 May-02 Feb-04
To Feb-89 Mar-90 Mar-93 Feb-94 Mar-95 Aug-95 Jun-96 Nov-97
Jul-00 Mar-01 Apr-03 Mar-04
4,370
women
15-54
4,595 9,925 5,040 4,925
7,070
women
and 1,996
men 15-54
5,515 6,655 10,773
7,246
women
and 1,962
men 15-54
9,711 18,000
9 1 1 1 4 4 4 3
Socioeconomic ¦ ¦ ¦ ¦ ¦ ¦ ¦ ¦
Labour force ¦
HH Enterprise ¦ ¦ ¦
Crops ¦ ¦ ¦
Woman ¦ ¦ ¦
Man ¦ ¦
Establishment ¦ ¦ ¦ ¦ ¦
Community ¦ ¦ ¦ ¦ ¦ ¦
Oct-89 Feb-91 Dec-93 Sep-95 Jun-97 Aug-96 Dec-97 Mar-99 Jan-01
Dec-01 Jan-03
Feb-94 Jun-96 Dec-97 Jul-98 May-00 Jan-02 Nov-03
Sep-94
8 11 9 19 27 12 18 16 6 9 -3
11 28 33 25 30 18 7
18
Survey
Que
stio
nnai
res
Duration
Date of Release of
publications
Sample Size
(usable observations)
Districts Ommited
Months from completion
to publications
-
The African Statistical Journal, Volume 3, November 2006 119
The IHS was a comprehensive survey covering a 12 months period
and provided the first ever baseline poverty estimate for the
country.
The subsequent four rounds of Monitoring Surveys (MS1-MS4) were
conducted with the main objective of providing time series data to
measure economic growth and social development. Although monitoring
surveys covered the whole country, they were much smaller in sample
size (ranging between 4900-6500 households) and covered a period of
between 7-10 months. Both the IHS and the First Monitoring Survey
(FMS) yielded useful data on small and household based
non-agricultural enterprise activities critical in updating the
National Accounts. The Second and Third Monitoring Surveys (TMS) in
addition to the socio-economic module included an agricultural
survey as the core-sub-ject module while the Fourth Monitoring
Survey (National Household Survey 1997) cov-ered a labour force
component as the core module as well as socio-demographic data. The
subsequent national household expenditure surveys conducted in
1999/2000 and 2002/2003, covered agriculture and labour force
modules respectively. The expendi-ture component of the
socio-economic module was maintained in all surveys except the
demographic surveys to support the derivation of poverty
estimates.
The Bureau is currently implementing the 2005/06 Uganda National
Household Survey. The survey focuses on socio-economic
characteristics, agriculture and relat-ed characteristics among
other modules.
5.2 Demographic and Health Survey
The Demographic and Health Survey (DHS) was conducted in 1989,
1995 and 2000/2001 respectively and collected data on fertility,
family planning, maternal and child health and other demographic
characteristics. In all, DHS information on HIV/AIDS was limited to
knowledge about awareness. Although there is widespread knowledge
about ways of transmission and protection, HIV/AIDS continues to
pose a serious challenge to the health delivery system and poverty
eradication efforts in Uganda. As a mechanism of monitoring the
incidence and prevalence of the pan-demic, a national-wide HIV/AIDS
Sero Survey was undertaken by the Ministry of Health in
collaboration with UBOS to, among other uses, provide prevalence
rates of HIV/AIDS in Uganda. The survey, unlike in many other
countries in eastern and southern Africa, was undertaken
independent of the DHS.
Evidence clearly shows that UBOS has overtime, evolved into an
established organi-zation with proven capability and expertise to
conduct a wide variety of nation-wide
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006120
household surveys. The Bureau has succeeded in organising and
conducting regu-lar surveys and consistently reduced the time gaps
between the end of field operation and the release of publications.
Most importantly, the organization has developed and maintained
expertise and capabilities in the design and implementation of
household surveys, which is obviously an essential asset for future
household survey system.
5.3 Survey organization and participation of stakeholders in the
development of household survey instruments
The Bureau does not have regional or district based offices
which makes recruitment of field based interviewers during data
collection very difficult. A mobile team approach has been employed
for field organization and management of most of the household
surveys. The teams constitute centrally recruited and trained field
staff responsible for data collection. The field team recruitment
criteria is based on the education level, local languages spoken
and the clusters to be covered per region. The team comprises of
five field staff headed by a supervisor and composed of four
interviewers, who travel to-gether from one survey cluster to the
next throughout the data collection period.
The alternative to the field team approach is the field based
interviewers in which in-formation is collected within or near
their area of residence. This model has not been applied in Uganda
because the Bureau does not have fully operational district or
re-gional offices. This makes the recruitment, supervision and
monitoring of field ac-tivities equally costly. Plans are underway
to operationalise this model once the re-gional statistical offices
are equipped and become functional. Although experience shows that
centrally recruited teams are expensive and unsustainable in the
long run due to the limited resources allocated to data production
by national governments, in the absence of operational regional
offices, the costs of decentralized field activities may exceed
those of the centrally recruited teams. A study is planned to
establish the costs and benefits of each approach as well as the
actual savings realized using the decentralized model of data
collection.
6. The Design of the National Household Surveys
6.1 Consultations with stakeholders
Household survey data has increasingly become more of a public
good than previ-
-
The African Statistical Journal, Volume 3, November 2006 121
ously imagined. Today, users demand for data rather than
indicators. For this reason, key stakeholders are involved at the
planning stage to ensure the production of de-mand-driven data from
these surveys. The wide involvement of different stakehold-ers
results into:
a) optimal utilization of household survey data for indepth
research by analytical and research institutions in Uganda;
b) generation of new ideas and information to enrich and support
evidence-based policy formulation and decision-making process;
c) promotion and enhanced multiple ownership of the household
surveys data thereby increasing data utilization in the development
planning processes; and
d) greater appreciation of the constraints associated with too
many demands on the questionnaire content to meet the diverse needs
of stakeholders.
Wide consultations have been made involving all the key
stakeholders in the produc-tion and use of data during the post
PEAP period. The PEAP was used as the build-ing block for the
survey programme and indicators identified as critical in informing
the PEAP were used as benchmarks.
6.2 Sample design
Sample selection and sample size
A similar sample design has been adopted for all household
surveys starting with the Integrated Household Survey (IHS)
1992/93. It is typically a stratified two-stage sampling design
except in some districts where the sample was selected in three
stages due to lack of Enumeration Area (EA) frames. At the first
stage, Enumera-tion Areas (EAs) are drawn with a Probability
Proportional to Size (PPS), while at the second stage; households
which constitute the Ultimate Sampling Units are drawn with Simple
Random Sampling (SRS). Stratification is done in such a way that
en-ables disaggregation of data to rural/urban and regional levels.
The demand for dis-trict and sub-district level indicators has
resulted in increases in sample sizes to de-rive some district
level estimates. The sampling frames used for all the surveys were
based on the Population and Housing Census of 1991 and 2002.
Household sur-veys conducted prior to 1991 were based on available
administrative information.
The sampling frame for selection of First Stage Units (FSUs) was
the list of EAs with the number of households based on cartographic
work of the 1991 and 2002 Population
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006122
and Housing Censuses. For selection of the second stage units,
which are the house-holds, a listing exercise was undertaken in
selected EAs in all the household surveys.
In all cases except the UNHS 2005/06, each district was
considered a stratum and divided into rural and urban sub-strata.
The urban area was further sub-divided into district town and other
urban areas. This multifaceted stratification enabled a better
spread and representation of the sample, thereby increasing the
efficiency of the es-timates. As shown in the Table 3, the sample
sizes have varied from about 5000 to 18000 households. The
allocated sample was selected with probability proportion-al to
number of households.
Level of disaggregation of data from household surveys
Uganda adopted a decentralized planning policy as a way of
improving governance and service delivery at local government
level. The challenge remains for the nation-al statistical system
to provide data and indicators at the various levels within a
lim-ited resource envelope. Alternative options have been developed
to try and fill the data gaps for example, the poverty mapping
initiative of the World Bank which re-lates household based survey
data with the population and housing census informa-tion. This
option is an alternative way of addressing the demand for small
area sta-tistics which is increasingly becoming an important
planning input.
Panel data: the quest for monitoring changes over time
Demand for Panel data has become a common request by major
household survey data users. It is argued that panel data of living
standards addresses a number of critical requirements of the PEAP
and MDGs including; appreciation of the function-ing of labor
markets, child mortality and labor, determinants of and returns on
edu-cation, among other issues. Nationally representative panel
data on living standards can also inform policy makers on
strategies for addressing equity issues between households and
across geographical areas. For these reasons, panel data in
addi-tion to being an important tool for program evaluation,
because of its focus on pover-ty dynamics, substantially complement
the ongoing poverty incidence data collection especially from UNHS
series. Panel data would also be an important complement to the
cross-sectional data of the household expenditure surveys as the
latter is most suited to provide information on poverty incidence
and other PEAP monitoring indi-cators. The focus on dynamics that
is inherent in panel data would illuminate better the reasons for
change in the various indicators (including poverty).
-
The African Statistical Journal, Volume 3, November 2006 123
Despite all the enumerated benefits, panel data poses a number
of problems to sur-vey statisticians. Tracing of panel households
for example, after a long time is the greatest challenge faced by
data collectors in Uganda. This is particularly experi-enced in the
urban areas with high levels of mobility. Strong collaborative
linkages supported by some incentive mechanisms could be introduced
to ensure continued respondents’ cooperation. Incentives may take
different forms ranging from gifts like t-shirts, and calendars to
actual cash. The latter incentive was tried out during the 1993/94
household survey but later abandoned because of the problems
associat-ed with its administration. Cash incentives also tend to
raise respondents’ expecta-tions thereby affecting the outcomes of
the survey.
The Bureau collected data from panel households based on the
1992/93 household survey. This involved tracing the same
respondents interviewed earlier (in 1992) in 1993/1994, 1994/1995
and 1999/2000. The 1999/2000 panel data was collect-ed on
approximately 1300 households. The Government, development partners
and research institutions have used the panel data in studying the
poverty dynamics be-tween 1992 and 2000. There is a strong argument
for the revival of panel data col-lection in Uganda to enable
monitoring of various changes in the economy. Tracing of panel
household members focused on household heads and their spouses. It
is also difficult to establish the extent of bias introduced by
inclusion of a panel. How-ever, the Bureau is about to engage major
stakeholders in a discussion to find an am-icable solution to panel
data question.
7. Data Analysis, Dissemination and Use
7.1 Data analysis
Collaboration in data analysis
Data analysis is one of the fundamental processes in statistical
operations. Failure to analyse the collected data renders them
useless to policy makers and undermines the real advantages of
household survey data in general. Timely dissemination of the
household survey results is largely dependent upon the successful
finalization of the data processing activity. UBOS considers this
process as one of the key steps in any household survey. Since the
1999, UBOS produces the first volume of the house-hold survey
reports within six months after field work. It is the policy of the
Bureau to ensure that data is shared with all partners and policy
makers before undertaking an-
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006124
other round of household surveys. To meet the increasing and
dynamic demands by policy-makers and other users of household
survey data, UBOS has entered into col-laborative arrangements with
the Economic Policy Research Centre (EPRC) of Mak-erere University
and the Makerere University Institute for Social Research (MISR) to
further provide the detailed analytical reports that policy-makers
need. This arrange-ment maximizes the utilisation of the limited
capacity at UBOS and other NSOs in Africa in micro level data
analysis.
Through such collaboration, the UBOS and EPRC supported by the
World Bank, Rockefeller Foundation, International Livestock
Research Institute (ILRI) and the Department for International
Development (DFID) embarked on a poverty mapping project. This
innovation involved linking household level information from both
the population and housing censuses and household surveys to
provide poverty rates at the lowest levels of administration. This
activity is an attempt to meet some of the data demands for small
area statistics.
Linking qualitative and quantitative data for poverty
monitoring
Poverty is a multi-dimensional phenomenon which is understood
and interpreted in various dimensions by different individuals and
communities. The conventional ap-proaches for deriving poverty
estimates using household income/expenditure in-volves a detailed
series of methodological steps and procedures. There are, how-ever,
other approaches to poverty measurements that use other
determinants which cannot be quantified. The participatory
approaches use a set of methods to as-sess the community and
household welfare by involving the communities in identify-ing and
defining the poor. Each of the approaches has its own merits and
demerits. Uganda has embraced both approaches in the poverty
monitoring process in order to maximize the benefits from the
strengths of each approach.
It is generally agreed in Uganda that quantitative approaches
make aggregation pos-sible, provide results whose reliability is
measurable and allow simulation of differ-ent policy options. On
the other hand, qualitative methods comprehensively define poverty,
provide more insight into causal processes and produce more
accuracy and depth of information on certain questions. It should
be noted that each approach has its weaknesses and the poverty
monitoring process in Uganda acknowledges them. The interest in
Uganda is to build on the strength of each approach to ensure that
they play a complementary role2. This is achieved at two levels:
First, house-
-
The African Statistical Journal, Volume 3, November 2006 125
hold surveys are capable of providing the research questions to
the qualitative stud-ies and the UPPAP II experience clearly shows
that this is possible and achievable. Not only were research
questions developed, almost half of the 60 clusters covered under
UPPAP II were panel sites for the 1999/2000 household surveys.
Second-ly, the qualitative approaches are useful in answering
questions that may not be eas-ily answered through a household
survey. It thus answers the questions on causes of the event,
coping mechanisms and social capital for health, education,
agricul-ture, orphanhood, poverty, migration, etc. in more detail
than one would expect from a household survey. The quantitative
approaches are useful in providing the magni-tude of the event and
answers to questions like, how many are poor, where are they, etc.
A section was thus included in the household questionnaire to
capture such is-sues but also to provide a national perspective of
the issues emerging out of the qualitative study.
Data reliability
Improvements in reliability
Improvements in data management have also evolved over time.
Range and consis-tency checks have been included in the data-entry
program. The processing team uses MS-ACCESS to carry out more
intensive and thorough checks, while various software are used to
compute statistical errors. Table 4 showed that while the na-tional
level Coefficients of Variations (CV’s) for the major indicators
are generally be-low 5 percent, they are below 10 percent at the
regional level as per the survey ob-jectives.
Standard errors and coefficient of variations
The household surveys undertaken by the UBOS and her predecessor
Department of Statistics provide estimates that are precise at
national, regional and at rural- ur-ban levels. Estimates at
district level are sometimes provided for fairly large districts.
The sample estimates are subject to sampling errors. These arise
because the sam-ple of respondents selected is only one of the many
samples that could have been selected from the same population
using the same design and expected sample siz-es. Sampling errors
are measured in terms of standard errors and reflect the
vari-ability between all possible samples. The coefficient of
variation and the standard errors both provide us with precision
levels that guide data users in assessing for
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006126
themselves the usability of the estimates. Some of the selected
key indicators based on the 2002/03 household survey are shown in
the Table 4.
Table 4: Standard errors and coefficient of variations (UNHS
2002/3)
Variable Estimates Standard Error
Coefficient of Vari-ation (C.V.)
Number of Cases
Average Household ExpenditureUganda 136,461 4,476 3.28
9,711Urban 258,049 22,792 8.83 4,062Rural 111,412 2,972 2.67
5,649
Average Household SizeUganda 5.1 0.046 0.90 9,711Urban 4.1 0.07
1.94 4,062Rural 5.3 0.05 0.95 5,649
Enrolment ratio (%)Uganda 85.60 0.65 0.76 11.353Male 85.07 0.82
0.97 5,503Female 86.09 0.76 0.89 3,984
Literacy Rate (10 Above)%Uganda 68.72 0.64 0.93 31,066Male 75.83
0.64 0.85 14,767Female 62.23 0.81 1.31 16,299
Literacy Rate (18 Above)%Uganda 67.6 0.70 1.0 20,637Male 78.76
0.69 0.9 9,599Female 57.98 0.92 1.6 11,038
Percentage of persons Ill or Injured during last 30 daysUganda
28.34 0.40 1.4 50,510Male 27.11 0.47 1.8 24,500Female 29.50 0.48
1.6 26,008
Percentage of population that usually sleeps under Mosquito
NetUganda 10.7 0.51 4.8 50,510Male 10.1 0.50 4.9 24,500Female 11.3
0.57 5.0 26,008
Houses by Roofing Type (%) Iron Sheets 63.3 1.1 1.8
9,711Thatched 2.4 0.3 10.6 9,711
Houses by Wall Type (%)Bricks 50.7 1.2 2.4 9,711Mud/Poles 45.8
1.2 2.6 9,711
Houses Floor Type (%)Cement 24.0 0.9 3.7 9,711
-
The African Statistical Journal, Volume 3, November 2006 127
Earth 73.5 0.9 1.3 9,711Concrete/Stone 2.0 0.2 10.0 9,711
Households by source of Fuel for Cooking (%)Electricity 0.5 0.1
26.0 9,711Paraffin 1.6 0.2 11.1 9,711Charcoal 18.0 0.8 4.5
9,711Wood 78.2 0.9 1.1 9,711
Household Ownership of (%):Bicycle 42.7 0.9 2.1 9,711Television
6.9 0.5 6.8 9,711Radio 63.3 0.9 1.4 9,711
It can be seen from the table that apart from variables like
concrete/stones, electric-ity and television sets which are not
common in all communities in the country, other variables have low
C.Vs of much les than 5 %. This shows that these estimates are
highly precise or reliable.
7.2 Data dissemination and use
Dissemination of data and information flow to users
Household survey data have provided critical inputs to policy
debate in Uganda. UBOS has a free and equal access policy to data
from both household surveys and other data sources. UBOS
information is released at the same time to both Government and the
general public. Household survey findings have been disseminated
using various channels. The main findings are disseminated at
national level through workshops and the media. Of recent, the
dissemination channels have widened to include Press Re-leases,
Publications, UBOS Website, Radio and Television and CD-ROMs.
How have the datasets been used in monitoring the PEAP?
The household surveys that UBOS conducts have been the major
source of data and information on poverty trends over a period of
10 years, an experience that is rare in sub Saharan Africa. In
addition, trends in mortality, fertility and other indica-tors have
also been generated through a series of demographic surveys. The
Bu-reau has deliberately encouraged researchers to utilise the
existing data ets to fur-ther inform policy dialogues and this has
gone down well with researchers in the
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006128
country. The production of the first ever poverty maps for
Uganda was a joint ef-fort between UBOS, Makerere University
Faculty of Economics and Management (FEMA), International Livestock
Research Institute (ILRI), World Bank and the UK Department for
International Development. Moreover, collaborative research
involv-ing UBOS staff and the research institutions has increased
tremendously over the last four years. The consistent increase in
the demand for the data is a manifesta-tion of the data utilization
phenomenon. Below are a few highlights on the data uti-lization of
UBOS data.
Not only has the data from household surveys enhanced trend
analysis, it has also in-formed several policy concerns including
the employment and unemployment rates, the nutritional status,
immunization coverage, and literacy levels and enrolments.
Ex-planations to some of the puzzles on the datasets have been
provided. Furthermore, Uganda committed itself to monitor the MDGs
and this is supported by the availabil-ity of regular and reliable
data. The household surveys have so far provided informa-tion to
monitor progress of some of the MDG indicators (poverty, youth
employment, maternal mortality, enrolment among others).
8. Concerns About Sustainability and the Long-term Household
Survey Programme
8.1 Financing of household surveys in Uganda.Nation-wide
household surveys are expensive undertakings that many governments
including that of Uganda are unable to finance out of their own
budgets. Household surveys in Uganda have mainly been funded by the
World Bank since 1988. Both the 1988 Household Budget Survey (HBS)
and the 1992/93 Integrated Household Survey (HIS) cost about US
dollars 900,000 and 2 million respectively. A major pro-portion of
the resources for household surveys were initially spent on
building ca-pacity and infrastructure development in form of
procurement of field vehicles and hiring of consultants. The two
surveys (HBS and IHS) were managed with techni-cal assistance from
UNDP. Thereafter, the more lighter monitoring surveys covering
about 5000 households were undertaken between 1993 through 1997 at
a cost of approximately US dollars 500,000. All these surveys were
undertaken on a regular basis but without a well documented program
of surveys.
2: Weaknesses of the quantitative approach include: (i) sampling
and non-sampling errors, (ii) misses what is not easily
quantifiable, (iii) fails to capture intra-household issues. For
qualitative methods, the weaknesses comprise of (i) difficulty to
generalize and, difficulties in verifying information
-
The African Statistical Journal, Volume 3, November 2006 129
However in 1998, a program of household surveys and censuses
covering the period 1999 through 2006 was developed and funded by
both the government of Uganda and development partners especially
the World Bank. Since then, government con-tribution towards
household surveys has been substantial. Government contribution
towards the UNHS 2005/06 for example was approximately US $
600,000. This is a substantial contribution and demonstrates
government commitment towards moni-toring the PEAP and MDGs. This
contribution was in form of project support to the Bureau, and did
not greatly affect other statistical programmes. Another programme
of household surveys and censuses has been developed and once
endorsed by the major producers and users of data, government would
be requested to commit fund-ing for the programme.
8.2 Concerns about sustainability
The PEAP monitoring and evaluation strategy identifies
indicators that are gener-ated from household survey data. The
progress made in monitoring poverty and the effects of other
interventions has been mainly based on the information from
household surveys. However, Uganda, like many developing countries
does not have enough resources to finance, on a sustainable basis,
regular household surveys and yet, these are invaluable sources of
data. The challenge is how to continue gener-ating data amidst the
constrained resource envelope. The household surveys un-dertaken by
the Bureau so far have been funded mainly by donors. Government has
also stepped up its financing of household surveys but because of
the resourc-es involved, it is unlikely that it would fully fund
the proposed household survey pro-gramme. A detailed costing plan
is being prepared to cover the survey programme and once finalized
further discussions will be held to enlist support from all the
stake-holders including development partners. Government has
indicated willingness to support the programme in line with the
PEAP objectives.
In addition to trying to convince government and development
partners about the need for making more ressources available to
continue the survey programme, the Bureau is carrying out a number
of innovations to ensure sustainability of the survey programme
including production of more innovative statistical products like
pover-ty maps, production of thematic and targeted reports,
speeding up data processing and analysis to achieve timelines and
is now looking into possibilities of establishing a permanent field
organization in form of regional statistical offices.
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006130
8.3 Future monitoring of PEAP for policy:–the long term
household survey programme
UBOS has developed a ten-year household survey programme that
will ensure con-tinuous monitoring of poverty indicators as well as
filling existing data gaps. The programme will continue to provide
nationally representative information for moni-toring the PEAP and
other development needs of Uganda. This is mainly driven by the
need for information at all levels of decision making. Figure 1
below outlines the proposed programme and the proposed periodicity.
The periodicity of the various household surveys has been
determined by the pace of the decision-making pro-cess. In Uganda,
this is given by the PEAP revisions every four years and by the
bi-annual Poverty Status Reports (PSRs).
Another factor considered was the dynamism of the indicators.
Some indicators – such as the rates of unemployment or immunization
coverage–reflect economic con-junctures or short-term government
actions that can be expected to change rapidly and may deserve to
be observed frequently. In contrast, other indicators, such as
ac-cess to electricity and housing condition among others, will
change slowly and may be measured adequately by the decennial
census or household surveys undertaken during the inter-censal
period. In between these two extremes, certain aspects of household
welfare – such as the amount and composition of household budgets,
and derived indicators such as poverty measures – may or may not
change quickly.
The level of geographic disaggregation was another aspect of the
household survey system that requires careful consideration, since
it is related to policy needs; it also has serious managerial and
budgetary implications.
According to the proposed household survey programme, surveys
with limited objec-tives use simpler questionnaires and pose fewer
exigencies in terms of the selection, training and supervision of
the field staff. Their simplicity gives them the potential of using
the large samples of households needed to provide sub-national
estimates, as well as national figures. On the other hand,
Consumption and Demographic surveys are much more delicate
operations. They use complex questionnaires that require careful
selection, training and supervision of field staff. As a result,
they are difficult to implement on very large samples and are
unable to provide results for numerous sub-national units.
3: The districts have now increased to 80
-
The African Statistical Journal, Volume 3, November 2006 131
4: Adopted from Munoz J: Household Survey Programme for Uganda
Bureau of Statistics 2004
Censuses
PSR
PSR
PSR
PSR
Population Census
PSR
PSR
PSR
Policy Framework
PEAP Revision
PEAP Revision
PEAP Revision
Household Surveys
Cartographic Updating
Cartographic Updating and Inventory of Infrastructure
ServiceDelivery Survey
ServiceDelivery Survey
DHS 4
Sero Survey
Sero Survey
Sero Survey
UNHS III
2012
2013
2014
2008
2009
2010
2011
2004
2005 Agricultural Census
2015 Agricultural Census
PEAP Revision
2006
2007
Census of Businesses and Enterprises (COBE)
2016
UNHS V
ServiceDelivery Survey
2017
2010
2011
DHS 6
UNHS IV
DHS 5
UNHS
2016
2017
2015
Sero Survey
2012
2013
2014
2004
2005
2006
2007
2008
2009
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
Figure 1: Suggested program of UBOS household surveys for the
period 2005-20171
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006132
Uganda presents a very particular challenge in this sense,
because the first level of geographic disaggregation below the
country is the district, and the country has 69 of them, some very
small. Data users have traditionally complained about UBOS’
dif-ficulties in providing district-level figures, pointing out to
the increasingly relevant role of districts in the decentralization
process.
Important as this policy factor might be with regard to the
demand for small area es-timates, the constraint imposed by
sampling theory are even harder. The fact is that the number of
households needed to make a good sample for a district is almost
in-dependent of the size of the district, and it needs to be fairly
large. However, very large samples, even in the unlikely case they
could be financially afforded, are vul-nerable to big non-sampling
errors, as a result of the difficulties of selecting, training and
supervising a tremendous group of interviewers.
In the figure the first column shows the time lines proposed for
each programme. The second column shows the national censuses and
other nation-wide data collec-tion activities that UBOS needs to
conduct in addition to household surveys. Some of these activities
are essential inputs into the household survey system, but even
those that may not be directly linked to household surveys still
need to be taken into account because they are big efforts that may
compete with household surveys for UBOS technical resources and
infrastructure (vehicles, etc.) These include; Agricul-tural
Censuses and the Population and Housing Censuses, updating the
cartograph-ic maps during the inter-censal periods and the Census
of Business Establishments (Uganda Business Inquiry).
The last column shows the proposed sequence of household
surveys. It consists of the cyclic replication of four elements
every four years: one Consumption Survey, one Service Delivery
Survey and two kinds of demographic and health surveys – a
“standard DHS” (with anthropometrics and no blood testing,) and a
“Sero Survey” (with blood testing and no anthropometrics.) As said
before, combining the last two elements into a single, integrated
instrument seems to be a technically feasible al-ternative that
deservers further exploration.
Consumption surveys need to be fielded over a 12-month period,
to take into ac-count the seasonality of the underlying phenomena.
The fact that all other surveys have traditionally been fielded by
UBOS over shorter periods is reflected in table 3. The proposal is
to spread the DHS surveys over the 12 months period as well to
cap-ture seasonality.
-
The African Statistical Journal, Volume 3, November 2006 133
Besides seasonality, another reason for spreading the data
collection period over several months or a year is to capitalize on
the heavy investment in human resources usually engaged in surveys.
It is cheaper and easier to select and train a small num-ber of
good interviewers and supervisors, engage then for many months,
than to train a larger number and use them for a shorter period.
Longer periods of data collec-tion also permit the implementation
of more reliable quality control measures. When the period is too
short, problems are often detected after the survey is finished and
no corrective actions are possible.
The implementation of the household survey programme, once
adopted will further ensure continued provision of policy relevant
information over a fairly long period of time
9. Conclusion
The poverty scourge in Uganda represents one of the greatest
challenges confront-ing all stakeholders in the development
process. The need for reliable, timely and regular information from
household surveys as well as other sources becomes even
increasingly more important than ever before. Policy makers need to
be informed of the progress (or lack of progress) in the targeted
sectors in order to address the identified constraints. It is
through the provision of reliable data that appropriate decisions
and interventions can be made. This is only possible if there are
regular mechanisms of data production and dissemination. The
long-term household survey programmes is aimed at sustaining the
data production process and hence keeping the development debate
and analysis alive.
References
Angus Deaton 1997: The Analysis of Household Surveys: A
Microeconomic Ap-proach to Development Policy
Juan Munoz and Kinnon Scott 2004: Household Surveys and the
Millennium Devel-opment Goals
Kakande M.: A Paper on Poverty Monitoring: The Practices and
Emerging Issues Lecture Notes April 2006
The Role of Household Surveys In Poverty Reduction Efforts: A
Case of the Uganda National Household Survey Programme
-
James Muwonge
Le Journal statistique africain, numéro 3, novembre 2006134
Ministry of Finance, Planning and Economic Development 2004:
Poverty Eradication Action Plan 2004/05-2007/08
Mijumbi P. and Okidi J. Analysis of Poor and Vulnerable Groups
in Uganda, Occa-sional Paper No 16
Ravi Kanbur, Cornell University: Qualitative and Quantitative
Poverty Appraisal: The State of Play and Some Questions January,
2001
Uganda Bureau of Statistics 2003: Uganda National Household
Survey Report: So-cio-economic Report
Uganda Bureau of Statistics 2001: Uganda National Household
Survey Report: So-cio-economic Report
Uganda Bureau of Statistics 2004: The draft Uganda Household
Survey System 2005-2015